4. Management Interface
Graphical User Interface (GUI)
Offers administrator full cluster control
Standalone desktop application
Manages multiple clusters simultaneously
Runs natively on Linux, Windows and MacOS
Cluster Management Shell (CMSH)
All GUI functionality also available through
Cluster Management Shell
Interactive and scriptable in batch mode
Cluster
Management
GUI
Cluster
Management
Shell
5.
6. Intel Phi Integration
Everything needed to enable MIC on a cluster is packaged
as easy-to-install Bright packages:
• MIC driver
• MIC runtime
• MIC SDK
• MIC OFED
• MIC flash utilities
Environment modules ensure that user environment is set
up perfectly (PATH, LD_LIBRARY_PATH, ...)
MIC driver recompiled automatically against running kernel
at boot-time
7. Intel Phi Integration
Set-up wizard takes care of initial MIC configuration
(e.g. creating bridge interfaces, assigning IP
addresses)
MIC appears as a first-class device type in cluster
management infrastructure
MIC can be configured, controlled and monitored
through CMSH and CMGUI
MIC is automatically added to the workload
management system as a consumable resource
Compute jobs may request MIC resource in job script
13. Intel Phi Workload Management
Three ways to run MIC jobs:
• Offload (i.e. MIC is used as accelerator from host)
• Native (i.e. job executes entirely on MIC)
• Symmetric (i.e. communicating processes on both host and
MIC)
Offload: MIC represented as consumable resource in
workload management system
Native: Ported Slurm to MIC
Symmetric: work in progress, will require some
changes to workload managers
Additional work in progress: make sure MIC is not
used in multiple modes simultaneously
14. Bright Cluster
Architecture — Monitoring
CMDaemon
metrics
data
Cluster
Management
GUI
Cluster
Management
Shell
Web-Based
User Portal
Third-Party
Applications
head node
node001
node003
node002
metrics
metrics
metrics
metrics
raw data consolidated
data
BMC
BMC
BMC
15.
16. Cluster Health Management
Goal: provide problem free environment for running jobs
Regular health checks
• Actions that return PASS, FAIL or UNKNOWN
• Can be associated with a settable severity and a message
• Can launch an action based on any response value
Pre-job health checks
16 MIC health checks included by default
Jobs will only be scheduled to nodes where MIC is working
properly (as determined by MIC health checks)
Intel Cluster Checker included to verify that cluster is set
up properly
17. Bright Cluster Manager makes it easy
to install, manage and use clusters
with Intel Xeon Phi Coprocessors.